Impact of CHD risk factors on related disability 1
The impact of risk factors for coronary heart disease on related disability in older Irish adults
Authors
Sharon M. Cruise, PhD1,2*
John Hughes, MSc2,3
Kathleen Bennett, PhD4
Anne Kouvonen, PhD2,5,6
Frank Kee, MD1,2
Affiliations
1 Centre for Public Health, School of Medicine, Dentistry and Biomedical Sciences, Queen’s University Belfast, Institute of Clinical Science B, Royal Victoria Hospital, Grosvenor Road, Belfast, BT12 6BJ, Northern Ireland, United Kingdom
2 UKCRC Centre of Excellence for Public Health (NI), Centre for Public Health, Queen’s University Belfast, Belfast, UK
3 Northern Ireland Statistics and Research Agency (NISRA), Belfast, UK
4 Department of Pharmacology and Therapeutics, Trinity Centre for Health Sciences, St James’s Hospital, Dublin, Ireland
5 Department of Social Research, University of Helsinki, Helsinki, Finland
6 SWPS University of Social Sciences and Humanities, Faculty in Wroclaw, Wroclaw, Poland
* Corresponding author: Dr Sharon Cruise, Centre for Public Health, Institute of Clinical Sciences Block B, Queen’s University Belfast, Royal Victoria Hospital, Grosvenor Road, Belfast, BT12 6BJ, Northern Ireland, United Kingdom. Email: ; Phone: 0044 (0)28 90971607; Fax: 0044 (0)28 90235900
Running head
Impact of CHD risk factors on related disability
The impact of risk factors for coronary heart disease on related disability in older Irish adults
Abstract
Objectives:To examine the prevalence of CHD-related disability (hereafter ‘disability’), and the impact of CHD risk factors on disabilityin older adultsin the Republic of Ireland (ROI) and Northern Ireland (NI). Methods:Population attributable fractions were calculated using risk factor relative risks anddisability prevalencederived from The Irish Longitudinal Study on Ageing and the Northern Ireland Health Survey. Results:Disability was significantly lower in ROI (4.1% vs 8.8%). Smoking and diabetes prevalence rates, andthe fraction of disability that could be attributed to smoking (ROI:6.6%;NI:6.1%), obesity (ROI:13.8%;NI:11.3%), and diabetes (ROI:6.2%;NI:7.2%), were comparable in both countries. Physical inactivity (31.3% vs 54.8%) and depression (10.2% vs 17.6%)were lower in ROI. Disability attributable to depression (ROI:16.3%;NI:25.2%) and physical inactivity (ROI:27.5%;NI:39.9%) was lower in ROI. Discussion: Country-specific similarities and differences in the prevalence of disability and associated risk factorswill inform public health and social care policy in both countries.
Keywords: coronary heart disease, disability, risk factors, relative risk, population attributable fractions (PAFs)
The impact of risk factors for coronary heart disease on related disability in older Irish adults
Introduction
The island of Ireland has seen a decline in mortality from coronary heart disease (CHD)(Bennettet al., 2006; Bennett, Hughes, Jennings, Kee, & Shelley, 2013);however, Irish CHD mortality rates are still amongst the highest in Europe (Bennett et al., 2006; European Health for All; Levi et al., 2009). A recent report forecasts increases of 50% for the Republic of Ireland (ROI) and 30% for Northern Ireland (NI) between the years 2007 and 2020in the numbers of adults who will ever have CHD (Balanda, Barron, Fahy, & McLaughlin, 2010). These increases in CHD are thought to be a result of both increasing populations (in terms of general population growth) and larger proportions of those populations who are in older age groups. As CHD is one of the leading causes of disability in older adults (Ebrahim, Wannamethee, Whincup, Walker, & Shaper, 2000; Adamson, Lawlor, & Ebrahim, 2004;Oldridge & Stump, 2004), increasing prevalence of CHD represents a keyissue for public health and for health and social care services in a climate of limited health care resources.
One island, two healthcare systems
The island of Ireland presents a unique opportunity to examine differentials in CHD prevalence and CHD-related disability. The one islandincorporates two countries, the Republic of Ireland (ROI) and Northern Ireland (NI) (the latter being a part of the United Kingdom), and although the two populations are similar in terms of ethnic and cultural background, diet, and lifestyle, each country has an independent health and social care system: the ROI’s is largely health insurance-based, but in some instances is a combination of public and private health services; while the majority of the population of NI is eligible tofree healthcare under the United Kingdom’s National Health System (NHS). There is mixed evidence for the impact of the differing healthcare systems on healthcare utilisation. For example, some studies (e.g., McGregor & O’Neill, 2007; Ward et al., 2009) have found that GP consultation and hospitalisation rates are much the same in both countries in spite of the availability of free healthcare in NI, while other studies (e.g., O’Reilly et al., 2007; Layte & Nolan, 2015) have found evidence of unmet need in some socioeconomic groups in ROI as a result of having to pay for GP appointments.
Risk factors for CHD and CHD-related disability
The associations between CHD and specific risk factors such as smoking, obesity, and physical inactivity are well established (World Health Organization, 2009; Yusuf, Reddy, Ôunpuu, & Anand, 2001a, 2001b). However, the literature that focuses specifically on risk factors for CHD-related disability is sparse. One of the few studies that considered the role of specific functional limitations after CHD onset, the Whitehall II study, found that of five lifestyle-related factors examined (obesity, smoking, alcohol, diet, physical inactivity), obesity and physical inactivity were most strongly associated with disability both pre- and post-onset of CHD (Britton,Brunner,Kivimaki, Shipley, 2012). If, as in the Whitehall study, we consider CHD as a mediator between various risk factors and subsequent disability,we can examine the effects of risk factors such as current smoking, obesity, physical inactivity, and diabetes (Yusufet al., 2001a, 2001b) onCHD-related disability. The effects of depression on CHD are more complex and the literature is inconsistent. However, there are a number of studies that have found depressive symptoms to be associated with the onset of symptoms of CHD (Hemingway & Marmot, 1999; WulsinSingal, 2003), andthe Global Burden of Disease Study (Charlson, Stapelberg, Baxter, & Whiteford, 2011) has flagged depression as a risk factor for CHD. Therefore, the present study will include depression as a risk factor for CHD and CHD-related disability.
Country-level differences in risk factors for CHD and CHD-related disability
Although there is a great deal of similarity between the populations of ROI and NI in relation to ethnic background, diet, culture, etc., previous studies have shown country differences in the prevalence of some of the risk factors for CHD. For example, Ward et al. (2009) found higher obesity levels in ROI’s 65+ population compared to NI, as well as higher smoking rates. However, Ward et al. (2009) found the NI population to be more sedentary than those in ROI.
NI has a long-established, higher prevalence of mental ill-health compared with the rest of the UK (O’Reilly & Browne, 2001). Furthermore, McGee et al. (2005) found that four times more people in NI (compared to ROI) scored in the clinically significant range for depression (as measured by instruments such as the CESD and the GHQ12, 8% vs 2%). The higher depression levelin NI is not unexpected – the well-documented ‘Troubles’ (a period of political conflict with accompanying civil unrest and violence that lasted from 1968 to 1998) are posited to have impacted on the psychological health of many in NI, especially those who lived (and still live) near contentious regions and peace walls (O’Reilly & Stevenson, 2003; Maguire et al., 2016). Although individuals living in the border areas of ROI (i.e., alongside the border with NI) during the period of the Troubles may have experienced some of this unrest and violence, the majority of the ROI population would not have been exposed.
Therefore, given the country-level variations in risk factor prevalenceshown in previous studies, it is reasonable to hypothesise some differences in how they may impact on CHD-related disability.
Socioeconomic differences as risk factors for CHD and CHD-related disability
Asocial gradient in cardiovascular morbidity and mortality is evident in most developed countries (Wilkinson & Marmot, 2003),and it hasbeen suggested that some (though not all) of the socioeconomic inequality in cardiovascular mortality and disability can be explained by a social gradient in conventional risk factors such as smoking and obesity(Beauchamp et al., 2010). Therefore, it is not unreasonable to anticipate some socioeconomic differentials in CHD prevalence, and in the impact of risk factors on CHD-related disability when stratified by socioeconomic position (SEP). Furthermore, differences in access to free healthcare between the two countries may also be an important determinant of CHD and CHD-related disability.
Therefore, theobjectives of the study were:i) to assess the extent to which disability associated with CHD varies by age, gender, and SEP in ROI and NI,;and ii) to compare the contribution of risk factors including smoking, diabetes, obesity, physical inactivity, and depression to CHD-related disability, stratified by age, gender, and SEP.
Methods
Samples
Information on CHD-related disability and risk factor prevalence, for estimation of relative risks,were sourced from high quality nationally representative studies.
The Irish Longitudinal Study on Ageing (TILDA)is a cohort study of ageing that is being carried out in ROI among a sample of more than 8,000 respondents aged 50 years and over. Detailed descriptions of the TILDA cohort, including study design and methodology,are described elsewhere (Kearney et al.,2011;Whelan & Savva, 2013). Further information is available at the TILDA website ( and the Irish Social Science Data Archive (ISSDA) website ( where the data are available on application. The present study used Wave 1 TILDA data which was collected between October 2009 and February 2011.
The Northern Ireland Health Survey (NIHS) is a cross-sectional population-based health survey that has been carried out annually in NI from 2010/11 among respondents aged 16 years and over. More information about the NIHS is available on the Northern Ireland Statistics and Research Agency (NISRA) website ( and on the UK Data Service website ( where the data are available on application. NIHS data used in the present study were collected during 2010/2011.
The response rate for both TILDA and the NIHS was 62%.
Pooled data
To provide more robust estimation of relative risks the datasets were merged to provide a pooled, all-Ireland sample after harmonisation ofall variables being used in the analyses.
Weighting
The TILDA and NIHS each have a population weighting variable that was applied to analyses involving the individual datasets in order to ensure that estimates were representative of the populations from which the samples had been drawn. TILDA weighting was based on age, gender and educational attainment; NIHS weighting was based on age and gender. It was not possible to apply the country-specific population weights to relative risk (RR) analyses involving the pooled dataset; however, all RR analyses were adjusted for gender, age, and SEP (i.e., the characteristics that are typically used to establish population weights).
Variables
CHD-related disability (disability)
In order to define CHD-related disability it was first necessary to establish prevalence of CHD. During the TILDA and NIHS computer-assisted personal interviews (CAPI), the respondent was shown a list of health conditions (which included ‘angina’ and ‘heart attack’) and asked to select any conditions that applied to them. In the present study, a respondent was deemed to have CHD if they indicated having had either angina or a heart attack.
The second step in defining CHD related disability was to establish prevalence of limiting long-term illness (LLTI). The LLTI questions in the TILDA and NIHS were broadly similar (see Table 1). In the present study, a respondent was deemed to have a LLTI if they responded ‘yes’ to questions 1 and 2.
Table 1.Questions used to derive limiting long-term illness (LLTI) in TILDA and NIHS
TILDA / NIHSSome people suffer from chronic or long-term health problems. By long-term we mean it has troubled you over a period of time or is likely to affect you over a period of time.
1. Do you have any long-term health problems, illness, disability or infirmity? NOTE:
INCLUDING MENTAL HEALTH PROBLEMS (yes/no)
2. Does this illness or disability limit your activities in any way? (yes/no) / 1. Do you have any long-standing illness, disability or infirmity? By “long-standing” I mean anything that has troubled you over a period of time or that is likely to affect you over a period of time? (yes/no)
2. Does this illness or disability limit your activities in any way? (yes/no)
Respondents were deemed to have CHD-related disability if they had both CHD and a LLTI. Hereafter, CHD-related disability will be referred to as disability.
Risk factors
Five established risk factors were included in the study and coding for these variables was standardised across the two datasets in order to facilitate merging of datasets. How each risk factor was defined is described below.
Smoking status (i.e., current smokers vs never smoked [reference]) and whether the respondent had diabetes (yes vs no [reference]) was derived from information provided during the CAPI (i.e., self-report) for both TILDA and NIHS.
Respondents’ body mass index (BMI)categorisations(derived from anthropometric measurement of weight and height in both TILDA andNIHS) were based on the World Health Organization’s (WHO) classifications of underweight (<18.5), normal weight (18.5-24.99 kg/m2), overweight (25-29.99 kg/m2), and obese (>30 kg/m2). In order to ensure adequate sample sizes in each category theunderweight and normal categories were aggregated into one category. This paper focuses on obesity versus the underweight/normal group.
Respondents were categorised as ‘physicallyinactive’ (low levels of physical activity) versus ‘physically active’ (moderate or high levels of physical activity) based on their responses to the International Physical Activity Questionnaire Short Form (IPAQ; Craiget al., 2003) which was administered during the CAPI for both TILDA and the NIHS. Note that although the IPAQ categories were available as a derived variable in the TILDA dataset, the meta-data did not make clear how it had been derived; therefore, we derived our own IPAQ categories using raw data in TILDA thus ensuring comparability with our treatment of the NIHS IPAQ data (using the authorised IPAQ scoring protocol – see
In TILDA, depression was assessed during the CAPI using the 20-item version of the Center for Epidemiologic Studies Depression Scale (CESD; Radloff, 1977). The CESD was designed to screen for depressive symptomatology during the seven days preceding assessment. A threshold of ≥16 on the total scale score is suggested as representing depression in the clinical range (Radloff, 1977). In the NIHS, depression was assessed using the General Health Questionnaire (GHQ12; Goldberg & Williams, 1988), a measure of common mental disorders for use in population studies. The GHQ12 was self-administered during the CAPI (i.e., there is a section of the NIHS CAPI where the interviewer hands the participant the computer and allows them to self-complete the more sensitive components of the questionnaire). A score of ≥4 on the total scale score has been suggested asan appropriate threshold to determine a mental disorder in the clinical range Mari & Williams, 1985). Respondents were classified as depressed(i.e., scores at or above the recommended threshold) versusnot depressed (i.e., scores below the recommended threshold).
Sociodemographic/socioeconomic variables
For the purposes of describing the age distribution of the sample, and estimating the prevalence of disability stratified by age, 10-year age bands were used (50-59; 60-69; 70-79; 80+). For the purposes of estimating relative risks (RRs) and population attributable fractions (PAFs), age was categorised as a dichotomous variable (50-64 and 65+). This was to maximise sample size/cell counts, and thus preserve power for RRestimation.
The present study used occupational group as an indicator of socioeconomic position (SEP). The NIHS used the National Statistics Socio-Economic Classification (NS-SEC) that is traditionally used by the UK’s Office for National Statistics (ONS); the occupational coding used in TILDA is similarto that used by Ireland’s Central Statistics Office (CSO)for the census. When deriving a SEP variable for our analyses, we had to ensure that the indicators of SEP between the two countries were comparable. The three SEP groups (professional/managerial [high]; lower non-manual [medium]; manual [low]) were broadly similar; however, in the NIHS there was a group of individuals coded as ‘no socioeconomic group (SEG), armed forces, etc.’ who were difficult to place. In TILDA there was a separate group for ‘farmers’ that was equally difficult to place. Excluding these two groups altogether or keeping them as separate SEP groups was not an option because of the effect this would have on sample/cell sizes (especially in the NIHS). Therefore, we made the decision to compare the distributions of these respective groups against the distributions of the manual SEP group using alternative indicators of SES (e.g., educational level, housing tenure, household income). For both the ‘farmers’ group in TILDA and the ‘no SEG’ group in the NIHS the distributions using alternative measures of SES were broadly similar to the distributions of the respective TILDA and NIHS manual SEP groups. Therefore, the decision was taken to include each of these two categories with the respective country-specific manual groups.
Additionally, TILDA included a sizeable ‘not applicable’ category (n=2,323). As 78% of this group were women, we thought it possible that they had never worked outside of the home and therefore could not be allocated to a specific occupational group. The decision was taken to treat this ‘not applicable’ group as a separate and independent group with no counterpart in the NIHS rather than try to absorb them within one of the three SEP categories. There was also a ‘missing/refused’ group in TILDA with quite large numbers (n=796) which was difficult to integrate into the 3-category SEP variable and which was kept as a separate SEP group. Therefore, within both health surveys we had a 3-category SEP indicator (high, medium, low) that was broadly similar and that allowed us to make meaningful comparisons and ultimately pool data, but within TILDA there were two additional groups (‘not applicable’; ‘missing/refused’) that were retained in order to maximise sample size.